The Effect of Uncertainty of Risks on Farmers’ Contractual Choice Behavior for Agricultural Productive Services: An Empirical Analysis from the Black Soil in Northeast China

Abstract

Regulating farmers’ choice of agricultural production service (APS) contracts can help maintain the stability of transactions, enhance agricultural production efficiency, protect farmers’ rights and welfare, stimulate the healthy and sustainable development of service organizations, and promote the economic benefits of agriculture. Under the uncertainty of risk (RU), farmers’ willingness to sign a regulated service contract after purchasing APS is a key factor in reaching an efficient and high-quality partnership. Based on the survey data of the black soil area in northeastern China, this study uses the Heckman two-stage model and analyzes it under the logical framework of APS purchases and contractual choice behavior (CCB) under the role of RU. The main findings are as follows. First, the RU has a significant “inducing” effect on farmers’ CCB, and the higher the RU, the more farmers tend to make formal written contracts; among them, each unit increase in business risk increases farmers’ choice of written contracts by 0.797 units, which is an important factor affecting CCB. Second, farmers’ personal trust level played a positive moderating role in influencing CCB. Compared to the low trust level group, farmers in the high trust level group had a diminished degree of influence on CCB, but the intensity was relatively weak. Third, farmers’ personal, family, and social characteristics all influence their behavioral decision-making processes. In order to reduce the risk level of cooperation between farmers and service providers, which can promote the standardization of cooperation contracts, this study suggests that the government should use the publication of manuals and other forms to carry out legal literacy, promote labor market information transparency, reduce service risks, and enhance the effectiveness of the linkage between farmers and service providers.

1. Introduction

Agriculture is the basic guarantee for national economic and social development. Food is one of the hot issues of continuous global concern, and food security is also an important guarantee for economic development and social stability [1,2,3,4]. With the gradual weakening of China’s urban–rural dual structure system and the gradual weakening of the mobility constraints of the population, the development of urbanization has attracted a large number of rural laborers, which has caused great disturbance to food production, and the sharp decrease in the number of laborers has triggered a decrease in agricultural production efficiency [5,6]. The structural imbalance of the labor force has gradually caused the phenomenon of non-agriculturalization and non-foodization of agricultural land to emerge, and food production is facing many challenges such as labor constraints, imbalance between supply and demand, fewer benefits, and cost increases [7,8]. China’s high-quality economic development urgently requires an integrated consideration of the trade-offs and coordination between urbanization and food security [9]. The APS industry has been developed as a new form of agricultural technology extension and as an organizational system arrangement for China to realize an effective connection between small-scale farmers and agricultural modernization, which is an important support for the development of modern agriculture [10,11,12,13]. In 2017, the Guidance on Accelerating the Development of APS Industry issued by the Ministry of Agriculture of China greatly affirmed the relevance of agricultural outsourcing and took the development of agricultural outsourcing services as the main service method to promote the upgrading of the APS industry and the development of moderate-scale operations. In 2020, China’s Ministry of Agriculture and Rural Affairs released the Plan for High-Quality Development of New Agricultural Business Subjects (2020–2022), which once again emphasizes the need to continuously enhance the development strength, operational vitality and driving capacity of APS subjects. The APS industry is a fusion of three major industries and is a labor substitution service in a separable state of agricultural production links. The development of this industry is conducive to solving the problem of urban–rural labor imbalance and can help improve agricultural production efficiency and resource utilization efficiency, which is an important driving force to promote sustainable agriculture [14,15,16]. Therefore, how to maintain the smooth and healthy development of the APS industry is a topic that needs to be discussed in depth.

After years of development and the government’s attention, the overall level of China’s APS industry has achieved substantial progress, but there are still problems such as an unbalanced business structure and a low level of management of service organizations [17,18]. To further improve the level of APS and agricultural modernization, it is necessary to examine and judge the situation in the context of the agricultural reality of specific regions. In a preliminary small-scale survey conducted in northeastern China, the group found that another problem in the development of APS is the varying degree of standardization of contract forms, unstable interest linkages and unclear division of responsibilities, which are not conducive to legal traceability [19]. There are frequent changes in cooperative partners, no fixed service providers, and short-term cooperation among farmer groups in the transaction process. This unstable cooperative relationship not only interferes with the service efficiency and the accumulation of cooperative reputation but also affects the time preference of service subjects and reduces the initiative of service innovation. At the same time, most contracts are missing in the cooperation process between farmers and service providers and are mainly agreed on orally, which leads to opportunistic behavior and increases the uncertainty of risks and potential transaction costs, including the cost of changing service providers, information costs, and even litigation costs [20]. A contract is an institutional arrangement for the way resources are transferred, specifying the relationship, rights and obligations of the parties to the transaction [21,22]. Oral agreements, which occur more frequently among farmers, have little binding power in this form of contract, often relying only on moral constraints, and it is difficult to delineate responsibilities and rights in the event of opportunistic behavior or natural disasters and other non-human causes that make the contract unenforceable, etc. [23]. In addition, the emergence of this conflict is a hidden expense that is also subordinate to the transaction costs incurred by farmers in the transaction process. In other words, the irregularity of transaction contracts causes an increase in farmers’ transaction costs, reduces the sustainability of the partnership, and hinders the quality development of APS. Therefore, how to guide farmers to sign contracts in a standardized manner and improve the efficiency of cooperation based on the purchase of APS by farmers has become a key and difficult point to vigorously promote the efficient development of the APS industry. By exploring the influence mechanism of farmers’ contractual choice behavior (CCB), this study can more accurately locate the influence effectiveness of RU in maintaining the stability of cooperative relationships, so as to better promote the development of agricultural scale and the healthy development of the APS industry.

Most of the research that has been conducted on farmers’ CCB revolves around its influencing factors and its impact on matters such as income [24]. Specifically, it can be divided into two aspects: on the one hand, the three main factors affecting farmers’ CCB, namely, farmers’ preference characteristics [25,26], personal characteristics [27], and family characteristics [28,29], and on the other hand, the subsequent effects of CCB, including farmers’ economic efficiency and production efficiency [30,31,32]. Das et al. analyzed the mechanism of action affecting the choice of land lease covenants by studying data from a sample of 4301 farmers in Bangladesh under a randomized experimental approach, and showed that the increase in credit access had a positive effect on CCB in terms of the number and scope of covenants [33]. Han et al. found significant effects of contract duration, conversion method, and confirmation factors on CCB based on data from a survey of 353 farmers’ land transfer contracts in Chengde, China [34]. From the existing literature, it can be observed that research on farmers’ CCB has formed a hot topic in the field of high-quality development of agricultural economy. The existing research results provide a complete research basis and reference value for this study; however, the following areas for further improvement still exist in the existing research. First, this research area has not focused precise attention on the emerging industry of APS. Existing studies mainly focus on farmer CCB in areas such as contract farming and agro-industrial chains, and there is still a lack of research on alternative machinery services in the agricultural production chain. Second, in terms of research content, most of the studies are centered on farmers’ preference traits and purchasing behavior of APS services but ignore what role transaction characteristics play in the second stage (CCB stage) of farmers’ decision-making process. Among the items that influence farmers’ CCB, RU is a key factor that has been overlooked. Therefore, there is still a lack of deeper exploration of RU issues in transaction characteristics. For example, what are the factors that influence farmers’ behavioral decisions when choosing a type of contract, and how do they ensure stable cooperative relationships? How does RU influence farmers’ CCB during the transaction process? How can RU be minimized, and what are its influencing mechanisms? There is little existing research on these specific, real-world issues and a lack of a systematic theoretical analysis framework. Third, in terms of research methodology, existing studies mostly use simple linear regression models, lacking further consideration of the counterfactual situation. Farmers’ CCB itself has self-selection characteristics, and CCB occurs on the basis that farmers’ purchases of services have already been reached; therefore, the issue of self-selection should be incorporated into the design of the econometric model.

In summary, RU, as a major important characteristic of the transaction process, has an important impact on the CCB of a limited rational group of farmers, and a regulated form of contract has a more sustained influence on farmers than an instantaneous occurrence of service purchase. CCB plays an important role in the scale efficiency. Improving farmers’ production efficiency, improving the long-term stable development of service providers, and clarifying the influence mechanism of farmers’ CCB under RU conditions are key issues that need to be addressed. Farmers’ CCB is an institutional arrangement formed spontaneously during the transaction process, implying uncertainties and potential transaction costs in the characteristics of the transaction and likewise being subject to interference from various external factors. Therefore, this study constructs a theoretical analytical framework for farmers’ APS purchase behavior and CCB based on transaction cost theory, aiming to elucidate the factors that influence farmers’ decisions and measure the extent of their influence. This study uses field visit data from 961 farmers in the black soil region of northeastern China to empirically test the theoretical analysis framework using the Heckman two-stage model. In this model with counterfactual analytical power, farmers’ decisions are considered to be influenced by a combination of transaction risk, business risk, and opportunity risk, while farmers’ trust levels create moderating effects in the process that are jointly influenced by individual, family, and social characteristics. The study has two contributions. Theoretically, it can more accurately represent the behavioral decision process of small-scale farmers and provide new research ideas, analytical frameworks, and methodological systems for the application of transaction cost theory and the thesis of how to advance the stable development of APS. At the practical level, this study can provide suggestions at the micro-level for reducing the risk of agribusiness cooperation and regulating the APS trading market. At the macro-level, this study can provide an empirical basis for promoting the vertical division of labor in agriculture, realizing the organic connection between small-scale farmers and modern agriculture, solving the difficult problem of urban–rural structural imbalance, improving agricultural production efficiency and resource utilization efficiency, and maintaining national food security.

2. Theoretical Analysis

Transaction cost theory is the core theory of the new institutional economics. There is no uniform standard for transaction costs in economics. Coase and Willianmson successively proposed and improved the theory and clarified the difference between transaction costs and production costs [35,36,37]. Based on the behavioral assumptions of limited rationality and opportunism, Willianmson proposed that RU is one of the key properties of transactions, specifically, that the effects of this unobservable uncertainty on economic behavior are widespread and pervasive and therefore require the design of appropriate decision processes to deal with them [38]. This RU is reflected in the three dimensions of transaction risk, operational risk, and opportunity risk in the process of APS trading, and this study focuses on the impact of this trading characteristic on farmers’ CCB. Current research in the application of transaction cost theory covers a wide range of interdisciplinary disciplines, such as regional economics, business management, political science, health and nutrition, and energy transformation [39,40,41,42,43]. Based on transaction cost theory, Jia delves into the impact of the presence of RU on cooperative performance, opportunism, and the importance of risk management in the residential energy retrofit sector in China [44]. Bahli further demonstrates the importance of risk issues by analyzing the RU in managing IT outsourcing practices based on transaction cost theory and using scenario simulation methods [45]. Existing studies have achieved many valuable academic results using transaction cost theory, which has laid a solid foundation for this paper at RU and CCB. Based on this foundation, this study constructs a two-stage theoretical analysis framework of farmers’, “APS purchase–CCB”, under the role of RU (Figure 1).

2.1. Theoretical Analysis of APS Purchase Stage

Farmers’ behavior in purchasing APS still follows the characteristics of limited rationality and the goal of pursuing economic benefits. In the decision-making process, farmers will consider their resource endowment comprehensively to further analyze the pros and cons of purchasing services [46]. The socioeconomic environment (external constraints) and the individual resource situation (internal drivers) associated with farmers’ service purchasing behavior are observable factors that directly influence the occurrence of behavior, acting as both constraints and stimuli [47]. This study synthesizes factors that have variability in the rural human situation, economic environment, and natural conditions of the black soil of northeastern China, which can be specifically divided into two areas according to the height of perspective. First, at the macro-level, the differences between regions still have some influence on the behavior of micro-subjects. We take Heilongjiang, Jilin, and Liaoning provinces within the black soil region of northeastern China as an example. Although the historical process and human background of the three provinces are not very different and their cultural heritage and folk customs are very similar, there are still differences in the economic development levels of the three provinces. According to the official data released by the three provincial statistical bureaus in 2020, the economic development and the agricultural industry vary greatly between regions: the contribution of agriculture to GDP growth in Heilongjiang province was 0.67 percentage points, while in Jilin province and Liaoning City, it was 0.2 and 0.3 percentage points, respectively. Therefore, the analysis by province can show the differential changes more clearly.

Second, at the micro-level, farmers’ individual resource characteristics also have an impact on their behavior, which is mainly divided into two major parts: individual characteristics and family characteristics. Farmers’ personal characteristics are key factors influencing their decisions and choices. In this study, farmers’ physical health and willingness to continue farming were chosen as the main indicators to measure. This is mainly due to the fact that farmers’ health is a factor that directly plays a role in service purchasing behavior. The labor force in normal condition is able to maintain labor tasks on the farm (excluding labor resources that are attracted by well-paying positions and give up farming activities), but due to the limited healthcare in rural areas, resources are still skewed towards urban areas, which leads to farmers’ health being a constant concern for the government and scholars. The poorer the health status of farmers, the more difficult it is to complete the labor work in the field, especially for small-scale farmers who are financially constrained to pay for the costs required to purchase agricultural machinery. Therefore, they prefer to outsource the labor-intensive agricultural production process to a service provider to complete the agricultural production operation more efficiently by paying a certain APS fee. In addition, farmers’ family characteristics, such as the amount of labor input, the degree of aging, and productive asset inputs, all have an impact on their service purchases. The more productive asset inputs farmers have, the weaker the demand for APS and also the ability to perform outsourcing services by themselves. It is worth noting that farmers’ social network relationships also have a critical influence on their decision-making behavior. A widespread culture of “kinship” prevails in rural areas of China, with high residential density within the village area, short spatial distances between neighbors, and close communication [48]. This communication adds more information to the farmers’ decision-making process, which in turn affects the final behavior.

2.2. Theoretical Analysis of CCB Stage

This study is more interested in the second stage of farmers’ CCB process than the stage of farmers’ APS purchase, mainly because of a combination of the following two reasons. First, the importance of CCB. Unlike the product outsourcing model in manufacturing, the agricultural production chain is less separable, and the links are tightly linked and difficult to divide. The agricultural production cycle is long and vulnerable to the rain and heat of the year, making it difficult to divide responsibilities. Therefore, APS, as a “factor of production” input, is inter-term in nature and does not rely solely on the point in time when the service is purchased to determine the expected benefits. In the long, variable, and complex series of transactions that occur after the purchase of a service, the choice of a standardized written contract can be of great help, which can be beneficial for the determination of liability and legal traceability. This makes it easier to maintain a stable cooperative relationship between farmers and service providers and to promote scale and organization. Second, the concept of RU is inextricably linked to the three major transaction characteristics in transaction cost theory, which, according the theory, has a significant impact on the changes in transaction costs and the arrangement of transactions [49,50]. Farmers’ CCB is essentially an arrangement of transaction patterns, in which RU plays a role that cannot be ignored. To further discuss the viewpoint of RU’s influence on farmers’ CCB, this study will use transaction cost theory to set the foundation and extend it, i.e., the concept of transaction characteristics is used as a base point to model the mechanism of RU’s influence on farmers’ CCB.

As shown in Figure 1, RU is manifested in three dimensions of transaction risk, business risk and opportunity risk after the purchase behavior is generated, which affects farmers’ CCB moderated by the level of personal trust. First, in a state where farmers are investing in factors of production (labor, land, fixed assets, etc.), the purchase of APS replaces part of the original labor force and reduces the labor burden of farmers’ families. Farmers purchasing APS are exposed to risks (including transactional, operational, and opportunity risks), and the greater the probability that such risks will be realized, the greater the impact on the service provider’s effort and hence the yield. For the farmer, their total production cost includes both transaction costs and hidden transaction costs from variable risks. Based on the utility maximization objective, the farmer and the service provider reach a stable cooperative relationship only when the marginal output value after the APS intervention is equal to the sum of production costs and transaction costs, i.e., when the risk probability is zero. When the RU is larger, the transaction costs are increased, and a new equilibrium point is formed. When farmers are not satisfied with the value of output at that equilibrium point but want to purchase services, farmers prefer a type of contract that can constrain risk, such as a formal contract that is binding on the service provider. The equilibrium point reaches the farmer’s requirement when the service provider increases the marginal output through the constraint of certain external forces. It can be deduced that farmers with higher RU are more likely to choose formal contracts to constrain the behavior of the service provider. Based on the above analysis, Hypothesis 1 is proposed:

Hypothesis 1.

RU has a positive impact on farmers’ CCB.

Second, farmers’ CCB, while influenced by RU, is also likely to be moderated by the level of farmers’ individual trust. In Chinese rural society, interpersonal trust is one of the core elements of the economic system, and the level of one’s trust is built on social relationships formed through long-term interactions [51,52]. Under the condition of information asymmetry, the biggest risk faced by farmers’ CCB may be the problem of moral hazard due to the existence of opportunistic tendency behavior, in which case, the trust between the contracting parties becomes a driving force to facilitate the implementation of the contract. The level of farmers’ individual trust reflects not only the relationship between individuals but also the relationship between individuals and the collective, organization, or institution. Some studies have shown that the level of trust reduces farmers’ transaction costs [53]. That is, farmers’ personal trust level plays a moderating effect in the disincentive effect from transaction costs, which in turn affect farmers’ choice behavior for the type of APS contract. Based on the above analysis, Hypothesis 2 is proposed:

Hypothesis 2.

The level of personal trust has a positive moderating effect on farmers’ CCB.

3. Materials and Methods

3.1. Research Area

To validate the two-stage analytical framework of APS Purchase–CCB, the empirical analysis of this study is based on survey data from rural families and village collectives in Heilongjiang Province (43°26′~53°33′ N, 121°11′~135°05′ E), Jilin Province (121°38′~131°19′ E, 40°50′~46°19′ N), and Liaoning Province (118°53′~125°46′ E, 38°43′~43°26′ N) in northeastern China. Most farmers in these three regions in the northern hemisphere are mainly growing corn in spring, which accounts for about 40% of the country’s total production and occupies a very important position in the country [54]. The main reasons for taking this study area include the following. First is the importance of the region. The three provinces in northeastern China are located in the typical black soil area, with precious black soil resources, natural soil development with a deep humus layer rich in organic matter and suitable for the growth of crops [55]. In addition, the topographic features of the region are flat and suitable for mechanical farming and the development of the APS industry; the climate is mild and humid, with a temperate continental monsoon climate spanning the warm, middle, and cold temperate zones, suitable for the growth of corn and the division of agricultural production links. Second is the state of economic development. The northeastern region experienced a glorious industrialization process in its early days due to its rich resource storage, and it is an old industrial base of China. However, with the depletion of resources and the iteration of light industries, the economic development of this region is in trouble, and the unbalanced structure of three industries and slow economic growth has attracted the attention of the government and academia [56]. The survey activities based on this region can better clarify the ideas to improve the development level of the APS industry and help to integrate the development of three industries and mobilize the vitality and healthy development of the agricultural economy.

3.2. Data Sources

In order to maximize the accessibility of the sample and the progress of conducting the research, this study was conducted in July–August 2018, when the research was conducted to avoid the busy farming period (spring sowing and autumn harvesting) and to guarantee the traffic running needed for the field survey because of good weather conditions. To verify the adequacy of the research preparation, the research team selected rural areas in Tieling City, Liaoning Province, to conduct a small-scale pre-survey in spring. In the formal research process, the family questionnaires distributed to the farmers mainly included basic information on the sample farmers’ family demographics, maize production and operation, input and output of each segment, service selection, and psychological preferences. Upholding rigor and scientific rigor, this study used a multi-stage stratified random sampling method to survey villages and families in six cities under Heilongjiang, Jilin, and Liaoning provinces: Harbin, Qiqihar, Suihua, Siping, Tieling, and Changchun (see Figure 2). One to three counties were randomly selected from each municipality, followed by four, six, or eight villages randomly selected from the counties, respectively. The research method used in this study was one-on-one offline interviews, which went deep inside farmers’ families and in the fields for sampling. The entire survey process upholds the principle of rigor. The investigators have undergone rigorous group training in advance to avoid the easy to generate “noise” questioning methods. The obtained dataset includes information from 982 families living in 11 counties, of which 21 families provided information that was still incomplete (non-response, missing answers, responses with logical errors before and after, etc.), so these families were removed from the sample database. Thus, the sample used in this study consisted of 961 families, with a questionnaire validity rate of 97.86%. Overall, the sample distribution was reasonable and could reflect the basic situation of farmers’ participation in APS in the three northeastern provinces; the basic characteristics of the sample are shown in

Table 1. Basic characteristics of the sample.

City	Total in City		Selected for Survey
	Counties	Towns	Counties	Towns	Farmers
Harbin	7	109	2	3	67
Qiqihar	8	68	2	4	160
Suihua	6	104	1	2	125
Siping	2	38	1	1	96
Changchun	7	75	4	7	300
Tieling	5	78	1	2	234
Total	35	373	11	19	982

.

3.3. Research Methodology

3.3.1. Heckman Two-Stage Model

To confirm the two-stage theoretical framework of “APS purchase–CCB” for farmers under the influence of RU, this study used Heckman’s two-stage regression method to conduct an econometric test. The choice of this regression model is mainly due to the self-selection problem of the sample, which refers to the bias of the estimated parameters due to unobservable variables. Only purely exogenous, random, and “noise-free” events that are generated in a random state do not introduce parameter bias in the regression equation. However, some variables with counterfactual characteristics are difficult to be sampled during field research. Assuming that farmers’ CCB occurs randomly, then traditional econometric models can be applied, but farmers’ CCB is built on the premise of purchasing services, which is the second stage of the behavioral process, and thus estimation using least squares in this case is error-prone. The function of the Heckman two-stage model is to try to correct the estimation bias caused by this bias and make the regression analysis more accurate and rigorous [57]. Existing research has used the Heckman two-stage model to address the problem of sample selection bias in several areas, such as Mitchell’s use of the method to correct the problem of sample selection error due to counterfactual question item design [58]. Dalango et al., on the other hand, used this approach to analyze the adoption of fertilizer technologies and the determinants of their intensification by small-scale farmers in southern Ethiopia [59].

In this study, farmers’ CCB is composed of two sequential decision processes: the first stage is farmers’ decision to purchase APS; the second stage is what type of contract is further chosen by farmers who have already purchased APS. If farmers do not choose APS, their preference and choice behavior for the contract type is not observed. It can be seen that farmers’ CCBs are subordinate and that there is a sample selection bias, which needs to be analyzed by applying the Heckman two-stage model. In this paper, farmers’ CCBs are fitted by the constructed Heckman two-stage model, and the sample selection model is constructed as follows:

$$\begin{array}{l}{y}_{1i}=X_{1i}\alpha +{\mu }_{1i}\\ y_{1i}=\{\begin{array}{c}1,\underset{}{}if\hspace{1em}{y}_{1i}^{*}>0\\ 0,\underset{}{}if\hspace{1em}{y}_{1i}^{*}\le 0\end{array}\end{array}$$

(1)

$$\begin{array}{l}{y}_{2i}=X_{2i}\beta +{\mu }_{2i}\\ y_{2i}=\{\begin{array}{c}c,\underset{}{}if\hspace{1em}{y}_{1i}=1\\ 0,\underset{}{}if\hspace{1em}{y}_{1i}=0\end{array}\end{array}$$

(2)

Equation (1) is the choice equation, and Equation (2) is the outcome equation. In these equations, y_1i and y_2i are the explanatory variables, which represent farmers’ behavior of “whether to buy” and “what type of contract to choose” for APS, respectively. y^*_1i represents the unobservable latent variable; c represents the type of contract chosen by farmers; X_1i and X_2i denote the independent variables affecting the process; α and β denote the parameters to be estimated; μ_1i and μ_2i denote the residual terms, assuming μ_1i~N (0, σ²) and μ_2i ~ N (0, 1), both obeying normal distribution; i denotes the ith sample farmer. The conditional expectation of farmer CCB is:

$$\begin{array}{l}E\left(y_{2i}|y_{2i}=c\right)=E\left(y_{2i}|y_{1i}^{*}>0\right)\\ =E\left(X_{2i}\beta +{\mu }_{2i}|X_{1i}\alpha +{\mu }_{1i}>0\right)\\ =E\left(X_{2i}\beta +{\mu }_{2i}|{\mu }_{1i}>-X_{1i}\alpha \right)\\ =X_{2i}\beta +E\left({\mu }_{2i}|{\mu }_{1i}>-X_{1i}\alpha \right)\\ =X_{2i}\beta +\rho {\sigma }_{{\mu }_2}\lambda \left(-X_{1i}\alpha \right)\end{array}$$

(3)

In Equation (3), λ (•) is the inverse Mills ratio function; ρ denotes the correlation coefficient between y₁ and y₂; ρ = 0 indicates that the selection process of y₁ does not affect y₂; ρ ≠ 0 indicates that the selection process of y₁ affects y₂ and there is a sample selection bias; and σ denotes the standard deviation.

3.3.2. Variable Selection

First, the explanatory variables set in this study are farmers’ CCBs and service purchasing behaviors. The explanatory variables include farmers’ transaction risk (TR), operational risk (MR) and opportunity risk (OR), and the familiarity of service recipients. The commitment of the service organization to benefits and the intensity of supervision during the service period are selected as proxy variables for these three risk types. Familiarity of the service provider is a binary variable that takes the value of 0 when the service provider is someone familiar to the farmer and 1 when the relationship between the service provider and the farmer is an unfamiliar one. The variable of whether the service organization commits to output benefits is a binary variable that takes the value of 0 when the service provider is unable to commit and takes the value of 1 otherwise. The intensity of supervision during service is a multicategorical variable that takes the value of 1 if the farmer regularly supervises the service provider, 2 if the farmer occasionally supervises, and 5 when the farmer does not supervise at all. It is worth noting that the indicators of personal trust level in this study were measured by a scale based on Xing’s research method, in which four questions were selected, namely, “how much you trust your relatives”, “how much you trust your neighbors”, “how much you trust your villagers”, and “how much you trust strangers”. Each question has a score from 1 to 5, which corresponds to very distrustful, distrustful, generally trustful, trustful, and very trustful. The final value of individual trust level was averaged by summing the scores of the above 4 questions [60], further divided into two groups of high-level trust and low-level trust, and grouped to measure the moderating effect of trust in the CCB process.

Second, the control variables selected for this study include the aspects of individual farmer characteristics, family production and business characteristics, identification variables, and regional dummy variables. Among them, farmers’ factor endowment is the key variable to be examined, and this study interprets factor endowment mainly by the following indicators. First, farmers’ individual characteristics include their health status, willingness to work in agriculture, and professional skills. The second is farmers’ family characteristics, which mainly include the average amount of labor invested per mu, the degree of aging, investment in productive assets, the degree of road leveling, and social networks. The dummy variables were selected as province variables, with Heilongjiang province set to 1, Jilin province set to 2, and Liaoning province set to 3. In addition, to ensure the identifiability of the model, the ease of farmers to find APS in the region was selected as an instrumental variable. The reason for choosing this variable as an instrumental variable is that the degree of development of the regional APS industry has an important role in influencing farmers’ behavior in purchasing APS, but this variable does not directly affect farmers’ CCB. Considering the differences between provinces, this section also introduces a province dummy variable to control for this factor. The specific model variables descriptions and their statistical descriptions are shown in

Table 2. Descriptive statistics of the variables.

Variables	Definition	No. of Obs.	Mean	S.D.	Min	Max	Expected Signs
							APS Purchase	CCB
Dependent Variables
Contract Type	Oral Contract = 0; Written Contract = 1	961	0.25	0.43	0	1
Purchase Decision	No Service Purchase = 0; Service Purchase = 1	961	0.83	0.38	0	1
Uncertainty of Risk
Transaction Risk (TR)	Familiarity With the Service Provider Familiarity = 0; Unfamiliarity = 1	961	0.13	0.33	0	1		+
Management Risk (MR)	Service Provider Commitment to Expected Revenue No = 0; Yes = 1	961	0.22	0.41	0	1		+
Opportunity Risk (OR)	Frequency of Supervision: 1 = Frequently; 2 = Occasionally; 3 = Generally; 4 = Rarely; 5 = Never	961	2.52	1.88	0	5		+
Identifying Variables
Service Accessibility (IV)	0 = No Service Purchase; 1 = Very Difficult; 2 = Difficult; 3 = Average; 4 = Easy; 5 = Very Easy	961	3.59	1.15	0	5	+	+
Individual Characteristics
Health Level (IC1)	1 = Very Poor; 2 = Poor; 3 = Fair; 4 = Good; 5 = Very Good	961	3.91	0.93	1	5	−	−
Willingness to Farm (IC2)	1 = Very Reluctant; 2 = Unwilling; 3 = Fair; 4 = Willing; 5 = Very Willing	961	3.58	0.77	1	5	−	−
Professional Skills (IC3)	Mastery of Professional Skills: 0 = None; 1 = Mastery	961	0.20	0.40	0	1	+	−
Family Characteristics
Labor Input (FC1)	Average Mu of Labor Input (Work Days)	961	16.47	24.03	0	169	−	−
Degree of Aging (FC2)	Ratio of The Number of Persons Aged 55 or Older to The Total Number of Families	961	0.39	0.37	0	1	+	+
Productive Assets (FC3)	0 = No Farm Machinery; 1 = Farm Machinery and Self-use; 2 = Farm Machinery Half-leased and Half-used; 3 = Farm Machinery Idle	961	0.63	0.69	0	3	−	−
Levelness of Farmland (FC4)	1 = Very Poor; 2 = Poor; 3 = Fair; 4 = Good; 5 = Very Good	961	3.70	0.75	1	5	+	+
Social Networks (FC5)	Number of People Visiting Homes During the Holiday Season	961	2.16	0.84	1	3	+/−	+/−
Social Characteristics
Province Dummy Variables (SC)	1 = Heilongjiang Province; 2 = Jilin Province; 3 = Liaoning Province	961	1.86	0.76	1	3	+/−	+/−

.

4. Results

4.1. Descriptive Statistical Analysis

In this section, first, descriptive statistics for all variables included in the Heckman two-stage regression model are discussed. Second, a descriptive analysis of APS purchases and contract type choices of the sample of farmers surveyed in this study is presented. Third, the level of APS participation in each segment of agricultural production activities, such as agricultural material purchases or land tilling and farm maintenance and management activities such as fertilizer application or pesticide spraying, is described with specificity. The descriptive statistical analysis in this section aims to provide more insight into the importance of the two main phenomena in the agricultural production areas examined in this study, namely, APS purchases and agricultural contractual selection, as well as relevant background information that can be used to interpret the results of the descriptive statistical analysis.

A summary of the variables included in the Heckman two-stage model in this study is shown by the statistics in Table 2. On average, 84.7% of the sample data of 796 farmers who purchased APSs chose to work with someone they were more familiar with, and only a small percentage of farmers had risk-taking preferences and were willing to work with strangers. It is interesting to note that even if the service provider is someone they know well, farmers still supervise during the transaction, and up to half of them still do so frequently. This suggests that even the intervention of a high trust level cannot control the opportunity risk, and it is still necessary to use measures such as supervision to curb the risk. From the sample statistics, farmers’ health level is generally good, with an average of 3.91, which exceeds the general health status; their willingness to work in agriculture is relatively high and exceeds the average, which indicates the strong vernacular culture in rural China. In addition, the elderly are more dependent and emotionally attached to the land than the young and are more reluctant to get off the land. The survey data show that fewer people have professional skills, and only one-fifth of them have specific skills. The average level of family labor input is very low, with the highest input reaching 169 days, which highlights the imbalance between urban and rural labor from the side and reflects the potential development of the APS industry. The level of arable land is good, reaching a medium to high level, and northeastern China is located in a plain area, which is suitable for the input of agricultural machinery and conducive to further scale development [61].

Table 3. Participation of interviewed farmers in APS.

Province	APS Status	Num. (n)	Prop. (%)	Contract Type	Num. (n)	Prop. (%)	Service Type	Num. (n)	Prop. (%)
Heilongjiang Province	Purchased	116	66.9	Verbal Agreement	163	46.4	pre-production stage a	156	44.5
							mid-production stage b	101	28.8
				Written Contract	72	53.6	post-production stage c	238	67.9
Jilin Province	Purchased	352	89.5	Verbal Agreement	317	80.6	pre-production stage	291	74.1
							mid-production stage	234	59.6
				Written Contract	35	19.4	post-production stage	277	70.5
Liaoning Province	Purchased	209	96.3	Verbal Agreement	80	40.6	pre-production stage	84	38.8
							mid-production stage	58	26.8
				Written Contract	129	59.4	post-production stage	59	27.2

^a The pre-production segment includes tilling and seeding segments; ^b Mid-production stage includes fertilizer application, spreading, irrigation, and mulching; ^c Late production includes harvesting.

summarizes the farmers’ participation, contractual choice, and service size of APS in the sample. In the summer of 2018, 82.80% of farmers in the sample in northeastern China reported purchasing APSs. From the survey of each province, the acceptance of APS by farmers in Liaoning province is better than the other two provinces (96.30%), which largely depends on the fact that Liaoning province has a better industrial base. Liaoning Province implemented agricultural industrialization earlier in China and has established stronger ties with farmers, which facilitates the development of the APS industry. Of 796 farmers that purchased APSs, nearly four-fifths of all farmers did not sign formal contractual instruments (70.30%), choosing to use verbal agreements to conclude contracts with service providers and complete the entire subsequent transaction process. In terms of the status of each province, Liaoning and Heilongjiang provinces have relatively better contract specification, with more than half of the farmers who purchased services signing a written contract. In terms of the type of service, farmers in each province have different levels of preference for different segments, but in general, they all show a strong interest in the tilling and sowing segments in the pre-production period and the harvesting segment in the post-production period. This is because these segments are more dependent on large- and medium-sized agricultural machinery, and farmers have difficulty in obtaining large machinery, so they outsource these segments to service providers for production operations.

4.2. Regression Model Results for the APS Purchase Stage

This study uses the Heckman two-stage model based on StataMP 16.0 statistical econometric software to estimate the logical theoretical framework of “APS purchase–CCB”. The first stage (service purchase stage) involves the application of a probabilistic model in which the dependent variable is the farmer’s APS purchase behavior. This variable is binary and has a value of 1 if the farmer purchases APS and 0 otherwise.

Table 4. First stage probit regression for predicting purchasing behavior.

Variables	Coefficient	Std. Err.	z	P > | z |
IC1	−0.037	0.0805207	−0.46	0.645
IC2	−0.208 **	0.0897083	−2.32	0.020
IC3	−0.519 ***	0.1518852	−3.41	0.001
FC1	−0.016 ***	0.0026301	−5.90	0.000
FC2	0.392 **	0.1890453	2.08	0.038
FC3	−0.356 ***	0.0978707	−3.64	0.000
FC4	0.066	0.0850241	0.77	0.440
FC5	0.066	0.0810903	0.82	0.414
SC	0.840 ***	0.1198664	7.00	0.000
IV	0.591 ***	0.0577289	10.24	0.000
Constant	−1.284 **	0.6132843	−2.09	0.036
Number of total observations				961
Number of censored observations				796
Log-likelihood test statistic				−233.94 ***
Pseudo-R2				0.47

Note: ** indicates p < 0.01, *** indicates p < 0.001.

shows the estimation results of the probit regression model for the purchase phase, where a series of indicators are used to measure the degree of fit of the model. The chi-square index of the model is 413.49, with a p-value well below the standard 0.05 level (p = 0.000) compared to a chi-square distribution with one degree of freedom. Thus, we have accurate and strong evidence to reject the original homogeneity hypothesis and accept the alternative hypothesis that heteroscedasticity is actually present in the residuals of our model for this regression. These estimates suggest that the explanatory variables used in the model are significant for the resulting equation.

According to the regression results in the first stage of the model, among all the influential factors affecting farmers’ APS purchase behavior, farmers’ personal characteristics are strongly associated with their behavior. Farmers with a relatively high willingness to continuously participate in agricultural production activities (IC2) are more likely to keep their contracted land in a family-based production and management model and do not actively seek out and purchase outsourcing services, while the idea of signing an APS is also weaker. Such a dependence on the land makes the older generation of farmers distinctly willing to work in agriculture and less willing to purchase services. Such findings suggest that the development of the APS market is to some extent constrained by the inherent “human–land emotion” in rural areas. The embedding of APS in the agro-industrial chain tends to increase farmers’ income through two paths: technological innovation and labor substitution, respectively. However, farmers with certain professional skills are independent in the application of technology, and the attractiveness of technological substitution brought by service providers is weakened. In addition, farmers prefer to do the farming themselves, reducing unnecessary service costs.

Regarding farmers’ family characteristics and social characteristics, the regression results showed that they were indeed strongly associated with farmers’ behavior. The labor input of farmer families (FC1) measures the number of days that people within the family are involved in agricultural production activities, which shows to some extent the labor capital within the family. The higher the amount of labor input by the farmer’s family, the richer the labor capital and the lower the demand for APS. In other words, one of the roles played by APS is the substitution of labor, and when farmers already have enough labor capital, the propensity for the service diminishes. In the case of families under the characteristics of fewer members and younger laborers going out to work, the weak labor force can hardly take up the productive activities in the field and has to resort to service providers for agricultural labor. Similarly, the degree of aging (FC2) and the family’s investment in productive assets (FC3) is reflected in the quality of labor force and the level of mechanization, respectively. The higher the value of the degree of aging, the lower the overall labor capacity of the family, the higher the stress of living and the more urgent the need for APS. This is due to the fact that service providers can both replace some of the labor-intensive aspects of the production process, freeing up family labor, both so that family members can provide better care to the elderly and to promote labor force participation in other jobs and expand the family’s income sources. Equally, the higher the level of mechanization, the less families need to hire service providers for mechanized operations and the lower the demand for technology-intensive services. Regarding farmers’ social characteristics, the heterogeneity of provinces was shown to have an impact on farmers’ behavior, which reflects the role of the external environment in stimulating individual decisions. In addition, the higher the accessibility (SC) of APS as an identifying factor in the first stage of behavior, the shorter the time spent by farmers searching for service providers and the lower the implicitly consumed transaction costs, which “induce” farmers to purchase APS.

4.3. Regression Results of the CCB Stage

4.3.1. Base Regression Model

The above study only considered the effect of control variables on service purchase behavior, but ignored the role of RU in the transaction process. Therefore, based on the regression test in the first stage, this study will continue to use 961 research data to further examine the effects of RU on service purchase behavior using “familiarity with the service provider”, “whether the service provider is committed to output benefits”, and “frequency of monitoring behavior during the transaction” as proxy variables for TR, MR, and OR. The effect of RU on farmers’ CCB was further examined using “familiarity with the service provider”, “commitment to output benefits by the service provider”, and “frequency of monitoring behavior in the transaction process” as proxies for TR, MR, and OR. In the second stage of CCB, we used the least squares estimation method for the analysis, as shown in

Table 5. Results of the second-stage selection estimation.

Variables	Coefficient	Std. Err.	t	P > | t |	St. Coef.
TR	0.098 ***	0.0246222	3.97	0.000	0.0753
MR	0.797 ***	0.0210032	37.93	0.000	0.7623
OR	0.013 ***	0.0047312	2.66	0.008	0.0549
IC1	−0.020 **	0.0088022	−2.33	0.020	−0.0443
IC2	−0.025 **	0.0106724	−2.37	0.018	−0.0453
IC3	0.016	0.0206428	0.78	0.438	0.0149
FC1	−0.001	0.0004298	−1.50	0.135	−0.0359
FC2	0.037	0.0226003	1.64	0.101	0.03178
FC3	−0.018	0.0127464	−1.38	0.169	−0.0280
FC4	−0.031 ***	0.0107999	−2.84	0.005	−0.0536
FC5	0.026 ***	0.0097661	2.71	0.007	0.0513
SC	0.023 *	0.0126122	1.81	0.070	0.0402
Mills Lambda	0.006	0.0199245	0.29	0.769	0.0081
Constant	0.215 ***	0.0689758	3.12	0.002	-

Note: * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

. The F-index of the model was 154.02, with a p-value well below the standard 0.05 level (p = 0.000) compared to a chi-square distribution with a degree of freedom of 1, which is a good fit. These estimates suggest that the three key explanatory variables used in the model, as well as the nine control variables, are significant for the resulting equation. The inverse Mills ratio was not significant in this study, indicating that the weak autoselectivity of the sample does not affect the results of the regression.

According to transaction cost theory, RU, as a trading characteristic, is closely linked to the arrangement of trading methods. According to the results of the second stage of the model regression, transaction risk (TR) has a significant positive effect on farmers’ choice of transaction contract type and is the main factor influencing farmers’ decision-making behavior, which is consistent with the a priori expectation (see Table 2). This indicates that the more familiar and knowledgeable the farmer is about the service provider, the more pronounced the choice of informal contracts is. Specifically, the closer the relationship between the farmer and the service provider, the higher the level of credibility and trust accumulated between the two and the more the farmer tends to choose the verbal agreement. On the contrary, if the relationship between the farmer and the service provider is an unfamiliar one and the two parties have not yet established a trust bond, limited rational actors will prefer to choose the formal contract with higher protection effectiveness for the matter of rights defense in case of disputes. At the same time, operational risk (MR) also has a positive and significant effect on farmers’ choice of contract type, a result that is also in line with expectations. The service provider’s commitment to post-trade returns makes the probability of default higher because the APS transaction process is longer in duration and more susceptible to uncontrollable events such as natural factors. Thus, for agricultural products that are “futures” in nature, promises need to be documented in writing, and verbal promises are less effective in legal terms. In addition to this, opportunity risk (OR) has a significant positive effect on farmers’ CCB. This indicates that the more farmers supervise the service providers, the more they tend to choose oral contracts. This is mainly because farmers are often in a passive position brought by information asymmetry when purchasing APS, and they are generally not highly educated. Social network relationships in rural China are relatively thick and have reached a certain level of psychological contracting, but such oral contracts are hardly legal in nature and are very risky. Farmers still need to engage in behaviors of a supervisory nature, but this has a small degree of influence on the choice of contract type. In order to better compare the order of the degree of influence of the uncertainty of the three types of risk on farmers’ behavior, this study calculated the standardized coefficients of the independent variables, and the results showed that business risk plays the largest effect in this CCB process. Based on the above analysis, Hypothesis 1 was verified.

Regarding the regression results of the control variables, similar to the first stage, farmers’ personal characteristics, family characteristics, and social characteristics all bring about an impact on the choice of the type of transaction contract to vary degrees. Interestingly, farmers’ health status (IC1) did not have a significant effect in the first stage but had a negative effect in the second stage, i.e., the worse the health of the farmers, the more inclined they were to enter into written contracts. This result can be explained by the fact that farmers did not pay much attention to their own health level when considering whether to purchase services and that, in the first stage, farmers were more inclined to measure their own production conditions. In the second stage, the impending long-term cooperation raises the concern about risk, and the poorer the health status, the more the farmers pay attention to the benefits that the service provider brings, and therefore are more inclined to sign formal contract terms to reduce the risk. In addition, farmers’ willingness to continue farming (IC2) still maintained a negative effect on farmers’ behavior in the second stage, indicating that farmers’ dependence on and love for the land remained one of the factors that hindered the acceptance of APS. Farmers’ family characteristics differed in the CCB stage. Both farmers’ degree of farmland leveling (FC4) and social network strength (FC5) had a significant effect on the choice of contract type, which was not shown in the first stage. This is due to the fact that the lesser the degree of undulation of the farmer’s land is, in terms of operability, the more favorable it is for the entry of large agricultural machinery. Moreover, the worse the level degree of the land, the less favorable it is for mechanized farming, which will deplete the life of the machinery to some extent. Therefore, the better the level degree of farmland, the weaker the farmers are in the perception of risk, and by virtue of the quality of farmland, the lower the probability of unintended losses and the less distrust of service providers. In contrast, social network strength, a measure of the breadth and depth of farmers’ social interactions, plays a positive role in farmers’ CCB process. This is due to the fact that the more farmers interact with others, the greater the amount of information they obtain, and the more rational the decisions they make in consideration of the risks.

4.3.2. Regression Results of Moderating Effects of Personal Trust Level

In order to further demonstrate the role of personal trust level in mediating contractual choice behavior in the transaction process, this study uses a five-dimensional Likert scale to measure personal trust level using four questions (corresponding to the four main circles of interpersonal relationships in rural areas): “How much do you trust your relatives”, “How much do you trust your neighbors”, “How much do you trust people in your village”, and “How much do you trust strangers”. Drawing on Dang’s measure, this study calculated the degree of trust level, q, of individuals by aggregating and averaging the values reported by farmers under these four topics [62]. It was also divided into two groups based on the median (3), i.e., the sample group with high trust level (q > 3) and the sample group with low trust level (q ≤ 3), by conducting a two-stage grouped regression for both groups, and then exploring how the level of individual trust, under the influence of uncertainty of risk, brings about an impact on farmers’ behavior (

Table 6. Regression results for the moderating effect of trust level group.

Variables	Coef.	Std. Err.	t	P > | t |	St. Coef.
Model 1
TR	0.237 ***	0.062736	3.78	0.000	0.1772
MR	0.819 ***	0.030181	27.14	0.000	0.7903
OR	0.030 **	0.012157	2.43	0.015	0.1249
N	428
Model 2
TR	0.120 **	0.050127	2.39	0.017	0.0947
MR	0.795 ***	0.029804	26.68	0.000	0.7549
OR	0.029 ***	0.009578	3.00	0.003	0.1287
N	533
Model 3
TR	0.103 ***	0.031688	3.24	0.001	0.0801
MR	0.779 ***	0.027284	28.54	0.000	0.7384
OR	0.015 **	0.006267	2.46	0.014	0.0661
N	642
Model 4
TR	0.081 **	0.038603	2.11	0.036	0.0611
MR	0.835 ***	0.032749	25.50	0.000	0.8126
OR	0.008	0.006916	1.20	0.231	0.0373
N	319

Note: ** indicates p < 0.01, *** indicates p < 0.001.

). Table 6 shows the estimation results and robustness tests of the Heckman second-stage regression model in the grouping case within the contract selection phase.

A total of four models’ regression results is included in Table 6, where Model 1 shows the analysis of regression results for the sample group with low trust level on CCB, and Model 2 shows the analysis of regression results for the sample group with high trust level on CCB. As can be seen from the results, the marginal effect of transaction risk (TR) decreases from 0.237 (model 1) to 0.120 (model 2), and the level of significance diminishes. This indicates that the level of personal trust plays a negative moderating role in the positive effect of transaction risk on farmers’ choice of transaction contract, i.e., the level of personal trust weakens farmers’ aversion to transaction risk and choosing verbal agreement in a high state of trust even though they are strangers to the service provider. In this case, the warning effect of transaction risk on farmers to enter into written contracts was less effective. Similarly, business risk (MR) had a positive and significant effect on farmers’ CCB in both trust level sample groups, but the coefficient of influence showed a significant decrease in the high trust level group. This indicates that the personal trust level plays a negative moderating role in the positive effect of business risk on farmers’ CCB, that is, farmers remain skeptical about the commitment of the service provider under the influence of the personal trust level. Against the standardized coefficients, among the three types of RU, business risk consistently maintains the most dominant influence strength. In addition to this, opportunity risk consistently has a positive and significant effect on CCB with a slight decrease in the sample farmers in both trust level groups, that is, low trust level has a greater effect on farmers’ CCB than the high trust level group. This suggests that the level of personal trust plays a weak moderating role. To further test the robustness of the measurement results, this study replaced the measure of trust level with a dummy variable to replace the five-dimensional scale to develop the analysis (0 = no trust; 1 = trust). Model 3 and Model 4 in Table 6 represent the results of the analysis for the no-trust group and the trust level group, respectively, and the results are similar to the analysis of Model 1 and Model 2, with robust regression results. Based on this, Hypothesis 2 was tested.

5. Discussion

5.1. Synthesis of Previous Studies

Existing studies on farmers’ CCB issues have focused more on the perspectives of contract farming, the marketing of agricultural products, and contracting and transfer of land [63,64]. For example, Rondhi used data from the Indonesian Livestock Farm Household Survey (ILFHS) to further study the factors influencing farmers’ contract farming, and the results showed that factors such as cooperative services and farmer associations brought about an impact on farmers’ contracting behavior [65]. Mao et al. analyzed the cleaning behavior in Chinese rural farming from the perspective of imperfect contracts and social trust and showed that this cooperative relationship further promotes cleaning behavior through contract farming participation [66]. Fuchigami et al. creatively developed a phased decision support system with specific cases from the Kanudos region to provide guidance to farmers on how to choose a contract acceptance strategy [67]. Compared with the existent literature, the present study has the following marginal contributions. First, the research perspective is more novel. From the perspective of APS, this study focuses on this emerging industry with great potential and explores the influence mechanism of farmers’ CCB on this basis. On the one hand, it is beneficial to the maintenance of farmers’ property rights, and on the other hand, it contributes to the healthy development of the APS industry. Second, the study highlights more prominently the influence of RU on farmers’ CCB and pays attention to the moderating role played by individual trust level in it. The decision-making process of CCB is systematically analyzed, the main factors influencing farmers’ choice behavior are identified, and the influence mechanisms between variables at different levels are revealed. In addition, focusing the study on contract choice breaks through the limitations of the existing literature on purchasing behavior only. Maintaining a good and stable cooperative relationship not only improves agricultural production efficiency, but also drives service providers to invest in technological innovation. Third, in terms of research methodology, this study integrates the two decision stages of “APS purchase–CCB” and the structure of transaction cost theory to build a logical framework and research path hypothesis, which can provide new research ideas for other related studies on APS.

5.2. Deficiencies of the Study

This study also has the following shortcomings. First, the research area of this study is the three major grain-producing provinces in the black soil region of northeastern China, which is typical and scientific, but still limited in scope. The southern part of China is also an important commercial grain base, and there are different types of maturity periods such as “three times in two years”, “twice in one year”, and “three times in one year”. There are different types of crop maturity periods. On the basis of the different growing periods of crops, there are differences in the transaction processes and arrangements between farmers and service providers, and the differences in family characteristics of RU and farmers are still questions that need to be explored in depth. Therefore, in the future, the study area can be expanded to form grouped controls to better clarify the factors affecting farmers’ contract type preferences. Second, there is still room for improvement in the design of variables. This study chose transaction cost theory as the underlying framework structure and selected three alternative variables for RU based on the concept of transaction characteristics. However, farmers’ contract-type decision making is a comprehensive behavior that condenses economic characteristics, psychological characteristics, and managerial characteristics, and an analysis using only one theory is still inadequate. Moreover, the uncertainty of risk is difficult to observe directly, and further learning is needed to express it more appropriately. Therefore, in the future, we can try to use multiple theories and combine the characteristics of different disciplines to explore the CCB of farmers in a deeper way. Third, the main direction of this study is to investigate the contractual relationship arising from the transaction process of farmers’ purchase of APS, which involves two main subjects, farmers and service organizations, and the government has certain intervention influence in it. However, due to the limitation of the research sample, this study was limited to farmers, and it was not possible to discuss the service organizations in more detail, which is a pity for this study. In the future, the authors will attempt to supplement this study with further detailed discussions on the development of transaction contracts and the APS industry, starting from various types of service organizations (e.g., cooperatives, agricultural enterprises, and individual agricultural machinery service providers).

5.3. Future Research Prospects

During the future research period, the following aspects can be further explored in more depth. First, the design and selection of the research area will be further enriched. In the next study, the research area will not be limited to northeastern China, but will be divided into temperate zones and crop-growing periods to explore farmers’ decision-making behavior and the development mechanism of the APS industry under different trading characteristics. Most of China is located in the northern temperate zone, and a small part is located in the tropics, which makes it easy to select the study area. In addition, the time factor will be designed into the research process in order to better optimize the research results. Due to the cross-sectional data, the observation of farmers’ behavior is not continuous. In the future, tracking research will be conducted to form panel data to achieve dynamic observation of farmers’ behavior. In this way, the rigor and scientific validity of the results will be improved. Second, different theories will be tried, and the design of variables will be optimized. Farmers’ decision-making behavior is a multidisciplinary research topic; this behavior can be explained by the economic approach of pursuing benefits, or the behavioral theory can be used to dissect the mechanism of behavior generation from a psychological perspective. Therefore, in future research, we should try to shift our thinking and integrate the characteristics of multiple disciplines to interpret and explain findings based on different theories. In the selection of variables, different research methods can be studied. In addition to the quantification of indicators, the method of randomized trials can be applied to divide farmers into experimental and control groups to observe how farmers’ decision-making behaviors change under different intervention conditions. This research approach breaks down the barriers that make it difficult to observe counterfactual phenomena and allows for a more scientific analysis. Third, the research perspective will be further expanded. Although the APS industry is an agriculture-based industry, it is still a service-based industry with complex and variable service processes and different types of actors involved. Therefore, in future research, we can innovatively look for the important nodes and critical factors that maintain the stability of this cooperative relationship from the perspectives of service providers and government. Using game models and system dynamics models, we can simulate the transaction process and clarify the ideas to promote the stable development of agricultural scale and APS.

6. Conclusions

This study constructs a two-stage (APS purchase–CCB) analytical framework. Based on survey data collected in the main grain-producing areas of three northeastern provinces, the Heckman two-stage model was used to study the influencing factors and path mechanisms of farmers’ CCB from three dimensions of RU. The main findings of this study are as follows. First, farmers’ service purchases and CCB are consistent with the analytical framework of transaction cost theory, and the type of contract as a transactional institutional arrangement, RU, is the main factor that influences the type of arrangement farmers choose. The positive effects of transaction risk (TP), operational risk (MP), and opportunity risk (OP) on farmers’ CCB reflect the “induced” effect, and these three RU categories are effective systems for influencing farmers’ choice of formal contracts. Second, among the factors affecting the CCB stage, farmers’ personal trust level has a positive moderating effect on CCB. High levels of trust weaken the strength of the influence of transaction risk, business risk, and opportunity risk on farmers’ choice of transaction contract type, but the effect of this moderating power on opportunity risk is weaker. Third, farmers’ health level (IC1), willingness to farm (IC2), road leveling (FC4), social network (FC5), and provincial dummy variables (SC) has the greatest impact on farmers’ CCB at the CCB stage. Fourth, farmers’ awareness of normative contracts is increasing and RU is more important, driven by the “APS purchase–CCB” mechanism. However, the current development of the APS industry remains unregulated, which also hinders farmers’ willingness to enter into formal contracts and maintain stable cooperative relationships.

Based on the above findings, this study can generate the following policy recommendations. First, the government should pay attention to the normative aspects of the service market. On the one hand, the government should focus on implementing the standardized operation of the APS market. In view of the development of the APS industry in northeastern China, there are still problems such as low entry barriers and lack of unified assessment standards. How to make service providers carry out agricultural services in a scientific and standardized manner and perform correct and rigorous processes and procedures is an important issue that needs to be addressed. On the other hand, the publicity and popularization of basic legal knowledge for farmers should be strengthened. At present, the overall level of education in rural areas is low due to the incomplete education resources and the lack of awareness of the importance of learning, and the awareness of the law and the protection of rights is not sufficient. The government should strengthen guidance and supervision to promote the general improvement in the standardization level of the contract.

Second, the government should control the RU issues arising from the transaction process, and the government should play a regulating and controlling role to further reduce the RU as the two major groups of farmers and service providers are involved in the APS transaction, and the government should set up an online or offline information platform to regularly summarize and update the information of service providers and market prices so as to establish the first barrier for service providers to enter the market. In this way, it can weaken the risk problem that farmers encounter when looking for service providers and the uncertainty that comes from working with strangers. At the same time, the government should notarize and supervise the transaction process, which can build a trustful link between farmers and service providers, reduce the psychological burden of farmers, and enhance their happiness and satisfaction. In addition, APS organizations can provide farmers with complete information and communication channels by building intermediaries based on acquaintances, thus curbing opportunism, reducing the uncertainty cost of behavioral decisions, effectively resolving transaction risks, and promoting farmers’ adoption of APS.

Third, the government should actively promote formal model contracts to enhance the standardization of APS contracts. It is found that although verbal agreements can save farmers’ costs in the short term in the transaction process, it is necessary to choose formal contracts to impose legal protection in order to establish a long-term and stable relationship between farmers and service providers, because informal contracts are difficult to effectively prevent opportunistic behavior tendencies and the risks they bring. Therefore, the government should actively promote formal contracts and further standardize their terms and formats. They can be designed in a targeted manner according to the different characteristics of each region and be open and transparent to prevent opportunistic behaviors caused by information asymmetry. This will stimulate APS organizations to optimize their management model, innovate technical tools, better improve their service capabilities, enhance agricultural production efficiency, and achieve a “win-win” model with farmer groups.